chimerabellerophoncommand.cpp

   1 /*
   2  *  chimerabellerophoncommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 4/1/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimerabellerophoncommand.h"
  11 #include "bellerophon.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ChimeraBellerophonCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none","none","none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pfilter("filter", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pfilter);
  18                 CommandParameter pcorrection("filter", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pcorrection);
  19                 CommandParameter pwindow("window", "Number", "", "0", "", "", "",false,false); parameters.push_back(pwindow);
  20                 CommandParameter pincrement("increment", "Number", "", "25", "", "", "",false,false); parameters.push_back(pincrement);
  21                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "ChimeraBellerophonCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string ChimeraBellerophonCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The chimera.bellerophon command reads a fastafile and creates list of potentially chimeric sequences.\n";
  39                 helpString += "The chimera.bellerophon command parameters are fasta, filter, correction, processors, window, increment. The fasta parameter is required, unless you have a valid current file.\n";
  40                 helpString += "The fasta parameter is required.  You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n";
  41                 helpString += "The filter parameter allows you to specify if you would like to apply a vertical and 50% soft filter, default=false. \n";
  42                 helpString += "The correction parameter allows you to put more emphasis on the distance between highly similar sequences and less emphasis on the differences between remote homologs, default=true.\n";
  43                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  44 #ifdef USE_MPI
  45                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  46 #endif
  47                 helpString += "The window parameter allows you to specify the window size for searching for chimeras, default is 1/4 sequence length. \n";
  48                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default is 25.\n";
  49                 helpString += "chimera.bellerophon(fasta=yourFastaFile, filter=yourFilter, correction=yourCorrection, processors=yourProcessors) \n";
  50                 helpString += "Example: chimera.bellerophon(fasta=AD.align, filter=True, correction=true, window=200) \n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "ChimeraBellerophonCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 ChimeraBellerophonCommand::ChimeraBellerophonCommand(){
  61         try {
  62                 abort = true; calledHelp = true;
  63                 setParameters();
  64                 vector<string> tempOutNames;
  65                 outputTypes["chimera"] = tempOutNames;
  66                 outputTypes["accnos"] = tempOutNames;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
  70                 exit(1);
  71         }
  72 }
  73 //***************************************************************************************************************
  74 ChimeraBellerophonCommand::ChimeraBellerophonCommand(string option)  {
  75         try {
  76                 abort = false; calledHelp = false;
  77
  78                 //allow user to run help
  79                 if(option == "help") { help(); abort = true; calledHelp = true; }
  80
  81                 else {
  82
  83                         vector<string> myArray = setParameters();
  84
  85                         OptionParser parser(option);
  86                         map<string,string> parameters = parser.getParameters();
  87
  88                         ValidParameters validParameter("chimera.bellerophon");
  89                         map<string,string>::iterator it;
  90
  91                         //check to make sure all parameters are valid for command
  92                         for (it = parameters.begin(); it != parameters.end(); it++) {
  93                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  94                         }
  95
  96                         //initialize outputTypes
  97                         vector<string> tempOutNames;
  98                         outputTypes["chimera"] = tempOutNames;
  99                         outputTypes["accnos"] = tempOutNames;
 100
 101                         //if the user changes the input directory command factory will send this info to us in the output parameter
 102                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 103                         if (inputDir == "not found"){   inputDir = "";          }
 104
 105                         fastafile = validParameter.validFile(parameters, "fasta", false);
 106                         if (fastafile == "not found") {
 107                                 //if there is a current fasta file, use it
 108                                 string filename = m->getFastaFile();
 109                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 110                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 111                         }else {
 112                                 m->splitAtDash(fastafile, fastaFileNames);
 113
 114                                 //go through files and make sure they are good, if not, then disregard them
 115                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 116
 117                                         bool ignore = false;
 118                                         if (fastaFileNames[i] == "current") {
 119                                                 fastaFileNames[i] = m->getFastaFile();
 120                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 121                                                 else {
 122                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 123                                                         //erase from file list
 124                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 125                                                         i--;
 126                                                 }
 127                                         }
 128
 129                                         if (!ignore) {
 130
 131                                                 if (inputDir != "") {
 132                                                         string path = m->hasPath(fastaFileNames[i]);
 133                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 135                                                 }
 136
 137                                                 int ableToOpen;
 138                                                 ifstream in;
 139
 140                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 141
 142                                                 //if you can't open it, try default location
 143                                                 if (ableToOpen == 1) {
 144                                                         if (m->getDefaultPath() != "") { //default path is set
 145                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 146                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 147                                                                 ifstream in2;
 148                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 149                                                                 in2.close();
 150                                                                 fastaFileNames[i] = tryPath;
 151                                                         }
 152                                                 }
 153
 154                                                 //if you can't open it, try default location
 155                                                 if (ableToOpen == 1) {
 156                                                         if (m->getOutputDir() != "") { //default path is set
 157                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 158                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 159                                                                 ifstream in2;
 160                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 161                                                                 in2.close();
 162                                                                 fastaFileNames[i] = tryPath;
 163                                                         }
 164                                                 }
 165
 166                                                 in.close();
 167
 168                                                 if (ableToOpen == 1) {
 169                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 170                                                         //erase from file list
 171                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 172                                                         i--;
 173                                                 }
 174                                         }
 175                                 }
 176
 177                                 //make sure there is at least one valid file left
 178                                 if (fastaFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 179                         }
 180
 181                         //if the user changes the output directory command factory will send this info to us in the output parameter
 182                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 183
 184                         string temp;
 185                         temp = validParameter.validFile(parameters, "filter", false);                   if (temp == "not found") { temp = "F"; }
 186                         filter = m->isTrue(temp);
 187
 188                         temp = validParameter.validFile(parameters, "correction", false);               if (temp == "not found") { temp = "T"; }
 189                         correction = m->isTrue(temp);
 190
 191                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 192                         m->setProcessors(temp);
 193                         convert(temp, processors);
 194
 195                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "0"; }
 196                         convert(temp, window);
 197
 198                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "25"; }
 199                         convert(temp, increment);
 200                 }
 201         }
 202         catch(exception& e) {
 203                 m->errorOut(e, "ChimeraBellerophonCommand", "ChimeraBellerophonCommand");
 204                 exit(1);
 205         }
 206 }
 207 //***************************************************************************************************************
 208 int ChimeraBellerophonCommand::execute(){
 209         try{
 210
 211                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 212
 213                 for (int i = 0; i < fastaFileNames.size(); i++) {
 214
 215                         m->mothurOut("Checking sequences from " + fastaFileNames[i] + " ..." ); m->mothurOutEndLine();
 216
 217                         int start = time(NULL);
 218
 219                         chimera = new Bellerophon(fastaFileNames[i], filter, correction, window, increment, processors, outputDir);
 220
 221                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[i]);  }//if user entered a file with a path then preserve it
 222                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) +  "bellerophon.chimeras";
 223                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[i])) + "bellerophon.accnos";
 224
 225                         chimera->getChimeras();
 226
 227                         if (m->control_pressed) { delete chimera; for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } outputTypes.clear(); return 0;        }
 228
 229                 #ifdef USE_MPI
 230                         MPI_File outMPI;
 231                         MPI_File outMPIAccnos;
 232
 233                         int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 234
 235                         char outFilename[1024];
 236                         strcpy(outFilename, accnosFileName.c_str());
 237
 238                         char FileName[1024];
 239                         strcpy(FileName, outputFileName.c_str());
 240
 241                         MPI_File_open(MPI_COMM_WORLD, FileName, outMode, MPI_INFO_NULL, &outMPI);  //comm, filename, mode, info, filepointer
 242                         MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 243
 244                         numSeqs = chimera->print(outMPI, outMPIAccnos, "");
 245
 246                         MPI_File_close(&outMPI);
 247                         MPI_File_close(&outMPIAccnos);
 248
 249                 #else
 250
 251                         ofstream out;
 252                         m->openOutputFile(outputFileName, out);
 253
 254                         ofstream out2;
 255                         m->openOutputFile(accnosFileName, out2);
 256
 257                         numSeqs = chimera->print(out, out2, "");
 258                         out.close();
 259                         out2.close();
 260
 261                 #endif
 262
 263                         if (m->control_pressed) { remove(accnosFileName.c_str()); remove(outputFileName.c_str()); for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } outputTypes.clear(); delete chimera;  return 0;       }
 264
 265                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine(); m->mothurOutEndLine();
 266
 267                         outputNames.push_back(outputFileName);  outputTypes["chimera"].push_back(outputFileName);
 268                         outputNames.push_back(accnosFileName);  outputTypes["accnos"].push_back(accnosFileName);
 269
 270                         delete chimera;
 271                 }
 272
 273                 //set accnos file as new current accnosfile
 274                 string current = "";
 275                 itTypes = outputTypes.find("accnos");
 276                 if (itTypes != outputTypes.end()) {
 277                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 278                 }
 279
 280                 m->mothurOutEndLine();
 281                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 282                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 283                 m->mothurOutEndLine();
 284
 285                 return 0;
 286
 287         }
 288         catch(exception& e) {
 289                 m->errorOut(e, "ChimeraBellerophonCommand", "execute");
 290                 exit(1);
 291         }
 292 }
 293 //**********************************************************************************************************************
 294